dork-hunter

name	dork-hunter
description	Creates a subagent to autonomously find books, leaks, and other content via Google Dorks and WebSearch. Invoke when user wants to research a topic, find leaks, or expand the knowledge base.
tags	["research","dorks","subagent","osint"]

Dork Hunter Subagent

Context

This skill orchestrates a research subagent to autonomously find books, leaked documents, and other hard-to-find content from the web using advanced search operators (Google Dorks). It integrates with the w5-extractor skill to deepen the repository's research.

Instructions

When the user asks to "hunt for [topic]", "find leaks on [topic]", or invokes the dork-hunter skill:

1. Generate Dorks (Subagent Execution):

   • Run the Python script to generate targeted Google Dorks for the topic using the gemini -y subagent.
   • Command: uv run tools/dork_hunter.py "[Topic]"
   • The script will output a list of highly specific search queries (e.g., filetype:pdf "Topic" confidential).

2. Execute Searches:

   • Use your available WebSearch tool to execute the top 3-5 generated dorks.
   • Analyze the search results to identify direct links to PDFs, books, documents, or high-value webpages.

3. Download & Organize:

   • Use RunCommand with wget, curl, or browser download to acquire the identified materials.
   • Save downloaded files (PDFs, raw text) into vault/Investigations/Dork_Hunter_Media/.
   • Ensure filenames are descriptive and sanitized.
   • For PDFs: Ingest using uv run tools/pdf_ingest_docling.py "<pdf_path>" "<title>" "<source_url>" to produce structure-preserving Markdown with auto-linked verbiage.

   After download and W5 extraction, run the verbiage snapshot and context extractor so the new material is included in trend and candidate outputs:

uv run tools/verbiage_timeseries.py
uv run python tools/verbiage_context.py

Review pipeline/verbiage_candidates.csv for newly surfaced terms before adding them to the verbiage DB.

4. W5 Extraction & Documentation:

   • For every source added, you MUST extract the full "W5" (Who, What, Where, When, Why) per the rules in llms.txt/RULES.md.
   • Use the predefined templates in vault/Templates/ for all new entries.
   • Create source nodes for the findings, embedding quotes or raw text exactly as they appear.

5. Update the Comprehensive How-To & Link Entities:

   • Open and review vault/Investigations/Comprehensive_How_To.md.
   • Synthesize the new findings, organizing the extracted methods, leaked data, and operational knowledge into actionable "how-to" steps.
   • Append links to the newly downloaded materials in the vault/Investigations/Dork_Hunter_Media/ directory.
   • Always cross-link entities (people, concepts) continuously without creating unlinked orphan files.